Context
The plastic injection moulding machine required continuous interventions as the output quality was decreasing over time. These interventions ranged from cleaning actions to mould changes and press overhauls.
Need
Maintain the output quality above the required level by anticipating the type of maintenance to be applied during the production process.
Solution
The press has been equipped with a variable speed drive that provides several types of data while the press is running (speed, torque, current, etc.). In addition, quality measurement is available through systematic control after each production batch. This data was used by DiagFit to provide a software sensor of the output quality based on the VSD. DiagFit then analysed the deviation of the specifications used to divert the virtual quality sensor to predict the type of maintenance to be applied to bring the quality back to its optimal level. The resulting normative maintenance recommendations assumed that the history of maintenance operations was recorded and passed on for data-based modelling.
Results
Quality prediction using the press speed profile has an accuracy of 3% in the segment between 80 and 100%. 8 different maintenance operations were separated into 4 clusters and the quality deviation profile towards one or the other cluster indicated the type of maintenance to be applied. The benefits are: more than 50% less rejects, 40% less press downtime.